Planning and Strategy
Start with the business goal, not the activity type
Every Target engagement begins with a measurable objective: lift conversion rate, increase revenue per visitor, reduce bounce, or grow a specific segment. The exam expects you to translate a stakeholder request into a hypothesis, a primary success metric, and one or more secondary metrics that guard against unintended harm. Know the difference between a primary metric (the decision criterion) and guardrail metrics (revenue per visitor, AOV, form abandonment) that can veto a win.
Choosing the activity type
Match the test design to the question. A/B tests compare discrete experiences; multivariate tests (MVT) decompose a page into regions and evaluate combinations, so they demand far more traffic per element. Experience targeting (XT) delivers a defined experience to a defined audience without a control comparison. Auto-Target and Auto-Allocate use machine learning to shift traffic toward winners. A frequent trap is choosing MVT when the goal is to compare two complete page designs - that is an A/B test.
Audience and traffic planning
Define the audience before the test, not after. Consider whether the audience is large enough to reach statistical significance within the campaign window, and whether segment overlap or exclusions distort assignment. Understand the sample size / duration relationship: longer tests with more traffic detect smaller lifts, but running a test too long exposes it to seasonality and novelty effects.
Worked rule of thumb
If a stakeholder asks for a 2% lift on a 3% conversion rate, the relative lift is small, so you need a large sample and should set expectations that the test may be inconclusive. Always state the minimum detectable effect up front.
Common traps
Confusing statistical significance with business significance; ignoring guardrail metrics; planning multiple simultaneous tests on the same page without a collision strategy.